DESIGN-ORIENTATED WIND ENGINEERING STUDIESNEW CHINA CENTRAL TELEVISION HEADQUARTERSBy: Jiming Xie (Rowan Williams Davies & Irwin Inc.) and Alex To (Ove Arup & Partners Hong Kong Ltd.)

CONSULTING ENGINEERS & SCIENTISTS

INTRODUCTIONThe new building of the China Central Television(CCTV) Headquarters has a very unique andcomplicated structural system. Its structuraldesign required comprehensive wind engineeringstudies to determine detailed wind loaddistributions for potentially critical load patterns.These load patterns include twist loadingbetween two main towers and vertical loading ofthe top links. To meet these design requirements,a high-frequency pressure integration testingprocedure was conducted in RWDI’s boundarylayer wind tunnel studies.

CCTV HEADQUARTERS BUILDINGWith its unusual structural system andarchitectural features, the new building of theChina Central Television (CCTV) Headquarterschallenged RWDI wind engineers and structuralengineers with its complex wind-resistant design.The new CCTV building consists of two towers,one of 51 stories and the other of 45 stories.These two towers are linked together at upperportions, creating a large overhead structureabove Story 37, as shown in Figure 1. The tower’slower portions, below Story 10, are also linked asa podium structure, but on the other side of thetop link. The building has very complicatedthree-dimensional dynamic properties.

DYNAMIC ANALYSISThe dynamic analysis revealed that the important modes of vibrations for wind response included two towersswaying in the same direction, two towers moving in different directions, and the top link flipping up and down.As such, not only the horizontal wind loads, but also the vertical loads acting on the top link and the differentialloads between the two towers (i.e., twist) are crucial for the CCTV structural design. There was a concern that thevertical loads on the top link might generate unfavorable overturning moments on the building base due to itslarge offset from the base center. For structural design purposes, it must be ensured that all the critical windeffects have been considered.

Figure 1: New China CentralTelevision (CCTV) Headquarters

Reputation Resources Results

ISSUE NO. 26

In other words, the objectives of wind engineering studieswere not only to find the maximum overall wind loads, butalso to determine the worst wind load distributions (i.e., loadpatterns) in association with various complicated load effects.For assessing the building’s serviceability, the magnitudes ofwind-induced building accelerations, including bothhorizontal motions and vertical motions, needed to bepredicted.

HIGH-FREQUENCY PRESSURE INTEGRATIONMETHODA high-frequency pressure integration (HFPI) test (Irwin 1995)was considered the most appropriate wind tunnel approachfor this project. The high-frequency pressure integrationapproach typically consists of the following procedures:

(1) Measure instantaneous wind pressures over theexterior building surface as a function of winddirection and wind speed.

(2) Determine the mean and non-resonance (background)components of overall loads by integrating theinstantaneous wind pressures over the entire buildingsurface with proper weighting factors.

(3) Determine the modal forces by integrating theinstantaneous wind pressures over the entire buildingsurface with the building’s modal deflections asweighting factors.

(4) Determine the wind-induced structural responses bycombining the measured wind forces with thecalculated structural dynamic properties.

For practical application, the HFPI method has beenincorporated with a procedure similar to finite-elementanalysis (Xie 2004). This approach was included in RWDI’sin-house software for pressure integration analysis (PIA).

As the first step of PIA, a structure is divided into a numberof elements. For a building portion, the typical elementconsists of several floors, as shown in Figure 2. Each elementis defined by the following five sets of properties:

(1) Inertial properties, i.e., the element mass and theelement mass moment of inertia;

(2) Geometric properties, i.e., the element dimensions andcoordinates;

(3) Dynamic properties, i.e., modal deflections of theelement for each mode of vibration;

(4) Load effect properties, i.e., influence factors of theelement to the considered load effects;

(5) Exterior force properties that consist of a set ofweighting factors for generating the total shears andmoments on the element.

For a sloped roof portion, the typical element is shown inFigure 3.

PIA was designed for general purposes yet it is applicable forvarious kinds of structural systems, such as buildings androofs. PIA provides great feasibility for considering varioustypes of load effects and automatically produces effectivestatic load distributions that reflect the worst-case loadeffects with wind directionality being included. Therefore,PIA is a structural design-orientated software that (1)provides the most useful information for wind-resistantdesign, and (2) provides the most effective interface betweenthe wind tunnel tests and the structural design practices.

In the application of PIA to the CCTV project, the structuralengineers’ requirements were considered in the PIAcustomized process. These requirements mainly determined:

(1) the format of loading, whether the loading was beingexpressed as point forces acting on the main structuralsystem or being expressed as the pressures over thestructural surfaces; and

(2) the critical load patterns to be considered for structuraldesign.

To assess the building’s serviceability, the magnitudes ofwind-induced building accelerations, including bothhorizontal motions and vertical motions, were also predicted.

Figure 3: Typical element for sloped roof section

Figure 2: Typical floor elements

Figure 5: Twist mode

Figure 7: CCTV wind tunnel model

Figure 6: Vertical mode

APPLICATION OF PRESSURE INTEGRATION METHODTO CCTVThe coordinate system and dynamic model of the CCTV’sstructural design was adopted for the wind engineeringstudy, so that the predicted wind loads could be readily usedby the structural engineers. This dynamic model consisted offour main axes, Tower 1 axis, Tower 2 axis, Top Link axis, andBottom Link axis, as shown in Figure 4. The elementproperties were calculated based on these axes.

The first 9 modes of vibration were considered in theanalysis. The first two modes represented the sway modes innortheast-southwest direction, and northwest-southeastdirection, respectively. The third mode showed a twistmotion between the two main towers. A significant verticalmotion of the top link was noticed in Modes 4 and 8. Thetwist mode and the vertical mode are illustrated in Figures 5and 6, respectively.

In total, 285 pressure taps were installed on the 1:500 scalemodel to simultaneously measure wind pressures duringwind tunnel testing. To consider the near-field terraineffects, a proximity model, which simulated the surroundingbuildings and structures in details within 600m radius fromthe site, was included in the wind tunnel model. The far-field terrain effects were simulated using spires androughness elements to duplicate the representative windprofile and turbulence properties in the area. The 1:500 scaleCCTV wind tunnel model is shown in Figure 7.

Figure 4: Coordinate of structural system

The critical load patterns that needed to be examined indetail were discussed and determined by the structuralengineers and the wind engineers from viewpoints ofstructural design as well as wind response. These loadpatterns included the worst overturning moments and shearsabout various principle axes, the worst differential loadsbetween the two towers, the worst torsional loads, theworst loads on the top links, and the worst loads on eachtower, etc. These load patterns were then expressed asequivalent influence factors. Since the purpose of the studywas to determine the worst cases for the given loadpatterns, it is not necessary to know the true values of theinfluence factors. Only the relative values, or the patterns ofthe influence factors, were needed. This is different from theclassic definition of influence factors. To ensure the givendistribution of the pressure taps be sufficient for measuringoverall structural wind loads, the pressure model wasmounted on a base balance and measured for the overallmean loads. Good agreements were achieved between thebase balance measurements and the pressure integrationmeasurements.

Tower 2 Top Link

Tower 1

Bottom Link

Y

ZX

North

RESULTSFor illustration, Figure 8 shows representative differentialload effects between the two towers as a function of winddirection for a 100-year wind speed. The figure indicatesthat the worst case of differential loading occurs at Northand West winds. Figure 9 shows the wind-induced verticalload effects acting on the top link as a function of winddirection for a 100-year wind speed. The worst winddirection for vertical loading on the top link is fromsouthwest, in association with maximum uplift.

Figures 8 and 9 are the loads based on a 100-year windspeed (28 m/s of 10 minute mean speed at 10 m height instandard open terrain), assuming this wind speed applies toall directions. To consider the wind directionalitydistributions, the determined wind effects as a function ofwind speed and wind direction was statistically combinedwith a Beijing wind climate model. The Beijing wind climatemodel was based on 25 years of hourly surface wind recordscompiled at the Beijing Meteorological Station. Theprevailing strong winds in Beijing are from north andnorthwest, which are not coincident with the worstdirections for load effects shown in Figures 8 and 9.Therefore, the wind directionality effects typically lead to areduction in the wind response.

For each load effect, the corresponding effective static loaddistribution was calculated and given at each building floorlevel about the four structural axes shown in Figure 4. It wasfound that the wind quasi-static pressures on the horizontalroof/soffit surface were the main contributions to thevertical loading, while the vertical dynamic loading due tothe top link motion was relatively small. This confirms thesuccessful structural design of controlling the verticaldynamic response of the top link.

The predicted horizontal and vertical peak accelerations atthe top occupied floor are given as a function of returnperiod. These accelerations are found to be within thegeneral acceptable range for human comfort.

This project received an award from the General Office ofCCTV New Site Construction & Development Program,Beijing, China. The support of Ove Arup & Partners HongKong Ltd, the structural engineers of the CCTV project, andthe comments received from Dr. Alex To of Ove Arup aregreatly appreciated.

In addition to the structural wind loading study, RWDI alsoconducted the following studies for the CCTV project,

(1) wind pressure study for cladding design;

(2) environmental wind study;

(3) snow loading and snow drifting assessment;

(4) sliding snow and ice assessment.

Rowan Williams Davies & Irwin Inc. (519) 823-1311 www.rwdi.com

RWDI Anemos Ltd.01582 470250 www.rwdi-anemos.com

Wind and Microclimate Services:• Acoustics, Noise & Vibration • Microclimate• Environmental Engineering • Regulatory Permitting• Hazard & Risk • Industrial Processes• Wind Engineering

Figure 8: Differential Load effects between two towers as afunction of wind direction

Figure 9: Wind-induced vertical load effects as a function of

wind speed

All images are courtesy of Ove Arup & Partners Hong Kong Ltd.